mlogit postestimation: "system limit exceeded - see manual r(1000)"

[Klaus Steitzel]

Hi everyone

I have an error message that I don't really know how to interpret.

I want to estimate a multinomial logit model and afterwards I would like to estimate average marginal effects.

This is basically how my data looks:

. tab y x

           |           x
         y |        No        Yes |     Total
-----------+----------------------+----------
     Cat 1 |       675        269 |       944
     Cat 2 |       816        202 |     1,018
     Cat 3 |     1,827        282 |     2,109
-----------+----------------------+----------
     Total |     3,318        753 |     4,071

Not a huge data set, reasonable distributions of values, not too many categories, I would think.

Running the model also causes no problems:

. mlogit y x

Iteration 0:  Log likelihood = -4177.7065  
Iteration 1:  Log likelihood = -4129.9641  
Iteration 2:  Log likelihood =  -4129.282  
Iteration 3:  Log likelihood = -4129.2818  

Multinomial logistic regression                         Number of obs =  4,071
                                                        LR chi2(2)    =  96.85
                                                        Prob > chi2   = 0.0000
Log likelihood = -4129.2818                             Pseudo R2     = 0.0116

------------------------------------------------------------------------------
           y | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
Cat_1        |
           x |   .9485219   .0963971     9.84   0.000     .7595871    1.137457
       _cons |  -.9957179   .0450425   -22.11   0.000       -1.084   -.9074362
-------------+----------------------------------------------------------------
Cat_2        |
           x |   .4723767   .1013382     4.66   0.000     .2737575    .6709959
       _cons |  -.8060162   .0421051   -19.14   0.000    -.8885406   -.7234918
-------------+----------------------------------------------------------------
Cat_3        |  (base outcome)
------------------------------------------------------------------------------

Then I'd like to have AME's and so I run -margins- right after -mlogit-:

. margins, dydx(x) predict(outcome(2))
system limit exceeded - see manual
r(1000);

How can that be?

My Stata version is

. about

StataNow/MP 18.5 for Windows (64-bit x86-64)
Revision 26 Feb 2025
Copyright 1985-2023 StataCorp LLC

Total physical memory:       16.00 GB
Available physical memory:    2.46 GB

Stata license: Unlimited-user 4-core network, expiring 14 Dec 2025

Thanks so much
KS

[John Mullahy]

For what it's worth I do not reproduce the error running v17.0 on my Mac.

. about

Stata/SE 17.0 for Mac (Intel 64-bit)
Revision 21 May 2024
Copyright 1985-2021 StataCorp LLC

Total physical memory: 8.01 GB

Stata license: Unlimited-user network, expiring 23 Oct 2025
Serial number: 401809302077
  Licensed to: John Mullahy
               UW-Madison

.
. tab2 y x

-> tabulation of y by x  

           |           x
         y |         0          1 |     Total
-----------+----------------------+----------
         1 |       675        269 |       944
         2 |       816        202 |     1,018
         3 |     1,827        282 |     2,109
-----------+----------------------+----------
     Total |     3,318        753 |     4,071

.
. mlogit y x

Iteration 0:   log likelihood = -4177.7065  
Iteration 1:   log likelihood = -4129.9641  
Iteration 2:   log likelihood =  -4129.282  
Iteration 3:   log likelihood = -4129.2818  

Multinomial logistic regression                         Number of obs =  4,071
                                                        LR chi2(2)    =  96.85
                                                        Prob > chi2   = 0.0000
Log likelihood = -4129.2818                             Pseudo R2     = 0.0116

------------------------------------------------------------------------------
           y | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
1            |
           x |   .9485219   .0963971     9.84   0.000     .7595871    1.137457
       _cons |  -.9957179   .0450425   -22.11   0.000       -1.084   -.9074362
-------------+----------------------------------------------------------------
2            |
           x |   .4723767   .1013382     4.66   0.000     .2737575    .6709959
       _cons |  -.8060162   .0421051   -19.14   0.000    -.8885406   -.7234918
-------------+----------------------------------------------------------------
3            |  (base outcome)
------------------------------------------------------------------------------

.
. margins, dydx(x)

Average marginal effects                                 Number of obs = 4,071
Model VCE: OIM

dy/dx wrt: x

1._predict: Pr(y==1), predict(pr outcome(1))
2._predict: Pr(y==2), predict(pr outcome(2))
3._predict: Pr(y==3), predict(pr outcome(3))

------------------------------------------------------------------------------
             |            Delta-method
             |      dy/dx   std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
x            |
    _predict |
          1  |   .1379269   .0148045     9.32   0.000     .1089107    .1669431
          2  |   .0330583   .0170505     1.94   0.053    -.0003602    .0664767
          3  |  -.1709852   .0196763    -8.69   0.000      -.20955   -.1324204
------------------------------------------------------------------------------

. margins, dydx(x) predict(outcome(2))

Average marginal effects                                 Number of obs = 4,071
Model VCE: OIM

Expression: Pr(y==2), predict(outcome(2))
dy/dx wrt:  x

------------------------------------------------------------------------------
             |            Delta-method
             |      dy/dx   std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
           x |   .0330583   .0170505     1.94   0.053    -.0003602    .0664767
------------------------------------------------------------------------------